This month, a company called Voice2Text Innovations filed patent infringement lawsuits against two voice over Internet protocol (VoIP) services, Phone.com and Vitelity. Voice2Text has no website or any other information online; its sole asset appears to be US Patent No. 8,914,003—unsurprisingly, a patent on using voice recognition to convert a voicemail into a text message. Equally unsurprisingly, it’s filed those suits in the Eastern District of Texas, whose patent-troll-friendly practices we’ve thoroughly documented on this blog. The Voice2Text patent is so absurd that we had to honor it with our Stupid Patent of the Month award.
Here’s the Voice2Text patent’s first claim:
1. A system, comprising:
a voice mail system receiving and storing a voice mail for a user;
a speech-to-text conversion module converting the voice mail from audio data into text data; and
a delivery module transmitting the text data in an email message in an email format to an email address associated with the user and in a Short Message Service (SMS) message in an SMS format to an SMS address associated with the user.
It goes on to detail the general process of receiving a voicemail, using a speech recognition “module” to convert the audio to text, and delivering the text as an email and as an SMS message. Keep in mind that it says nothing about how the voice-to-text module works; it’s merely a patent on the system for sending a voicemail as an email and a text message.
That is to say, it’s a patent on doing one of the most obvious things you’d want to do with a speech recognition algorithm.
As with other software patents, such as this previous Stupid Patent of the Month from Blue Spike, the Voice2Text patent leaves the hard work of actually writing code to the reader. Since the patent describes nothing beyond banal generalities (e.g. the system “may be configured to convert the voice mail into a text”) it provides no meaningful assistance over simply starting the project from scratch.
But there’s another fundamental problem with this patent: it claims an idea that wasn’t new. According to the law, a person isn’t entitled to a patent if the claimed invention already existed when the application was filed or would have been obvious to someone skilled in the relevant technology area. Reliable voice-to-text conversion has been a goal in the software industry since the 1950s. New uses of that technology—like the voicemail-to-text features that many VoIP providers offer—have been enabled by improvements in the speech-recognition algorithms themselves. It would be naïve to think that voicemail-to-text was ever held back because no one had thought of it yet. Companies were already offering voicemail-to-text services long before the Voice2Text patent came along. In fact, a lot of them were.
The Voice2Text patent was filed in June of 2006. Here’s a 2005 ZDNet article about a product essentially identical to what’s described in the patent. We even found a very similar service going all the way back to 2001. That’s just scratching the surface. If you want to join in the fun and look for more examples of this “invention,” try a custom search in Google to retrieve only results from May 2006 or before. All of this material that existed prior to the “invention” now claimed by Voice2Text is referred to as prior art. When reviewing a patent application, part of the Patent Office’s job is to look for relevant prior art.
That we were so easily able to find relevant, material prior art points out a serious flaw in Patent Office practices. Too often we see examiners at the Patent Office only looking for patent prior art. That is, only looking for existing patents that are for the same invention or are close enough that the invention is obvious. The Patent Office routinely ignores the vast amount of prior art in the real world that’s not patented, and so we get new patents on old products that have existed for years.
We’ve long noted that the Patent Office does a poor job searching for prior art when reviewing software patents. Software patents issue every week that don’t describe new inventions—most of the time, a simple Google search proves it. The result is a system that rewards those who write vague wish lists at the expense of those that do the hard work of developing commercially successful systems.